LIBRARYKilling EEG Noise: 50/60 Hz & the Right-Leg Drive.

The EEG signal is tiny: tens of microvolts, dropping to a few µV. The 50/60 Hz hum from mains wiring is not. The body and the electrode leads act as antennas, capacitively coupling to the power line, and the resulting common-mode voltage on the body can run from a few millivolts to a couple hundred, hundreds to thousands of times larger than the brain signal you're after. Win this fight and EEG works; lose it and you record the power grid.

Where the hum comes from

The coupling is mostly capacitive. The mains wiring in the walls sits at 120 or 230 V and radiates an electric field; your body and the high-impedance electrode leads form the far plate of a small capacitor across that field, so a tiny displacement current flows through the body and builds a common-mode voltage on it (Webster, 2009). Because the path is capacitive, two things make it worse: higher electrode impedance, where the same pickup current develops a bigger voltage, and longer unshielded leads, which give the field more area to catch. Every real defense below, common-mode rejection, the bias loop, low impedance, and shielding, aims at this one capacitive path.

Common-mode rejection and CMRR

The first line of defense is the differential amplifier: it amplifies the difference between two electrodes and rejects whatever is common to both, and the mains hum is (mostly) common to both. How well it does this is its common-mode rejection ratio (CMRR); good EEG amplifiers hit 100–110 dB. But CMRR alone isn't enough, and here's the subtle part: if the two electrodes have unequal skin impedance, the common-mode voltage divides differently across each input and shows up as a difference, converting rejected noise back into signal the amplifier faithfully amplifies. Imbalanced electrode impedance, more than the chip, is usually the real limit.

Shielding and guarding the leads

A shield is a conductor wrapped around the signal lead that intercepts the mains electric field before it reaches the wire inside. A plainly grounded shield works, but it adds capacitance from the lead to ground, and on a high-impedance EEG lead that stray capacitance bleeds signal and unbalances the two inputs. The fix is guarding: drive the shield with a buffered copy of the lead's own common-mode voltage so the wire and its shield sit at the same potential. With no voltage across the wire-to-shield capacitance, no current flows through it, so the cable stops loading the signal while still blocking the field (Webster, 2009). A driven shield and the driven-right-leg loop are the same trick applied in two places.

Capacitive (electric-field) pickup is the dominant mains problem on high-impedance bio leads, so shielding plus high CMRR cuts it hard, though never all the way to zero (Webster, 2009). Magnetic (changing-current) pickup is usually smaller for mains, but it grows near transformers and motors, and the defense there is a twisted pair, which shrinks the loop area the field threads through. Keep the leads short, twist them, and where the environment is brutal, put the subject and the amplifier inside a screened (Faraday) enclosure.

How to beat mains noise, in order

• High, balanced common-mode rejection: a good instrumentation amp AND matched electrode impedances.
• A driven-bias / right-leg loop to actively null the common-mode mains voltage.
• Low, STABLE electrode-skin impedance (prep the skin): a steady 50 kΩ beats a fluctuating 10 kΩ.
• Short, twisted leads and a driven shield to cut capacitive pickup and cable artifact.
• Distance from mains-powered equipment and power cables.

▸References

▸Keep going

Designing the front-end where all of this lives, the CMRR, the bias loop, the grounding and shielding, is the build in the OTD Academy EEG front-end project.